Stability

The highest frequency one always wins in real life.

Or they knew their audience.

John

Not necessarily. Three-integrator loops go unstable with large- amplitude, slow oscillations (AKA "wash machine mode") when the gain gets too small.

This is HILARIOUS! Of course, you find it in motion control; you also find "bang-bang " controllers, which are guaranteed to oscillate (but that doesn't mean they don't work well). The problem in amplifiers has to include the nasty little secret, of saturation. With a sufficiently large signal, or bias, ANY op amp might have unity gain at a wide range of frequencies, and it's not generaly prudent to ignore behavior away from unity-gain frequency.

Barkhausen's criterion is a blunt instrument, but a useful one.

You have to know why to use it, and when to ignore it. I once stopped a co-worker who was wiring in a compensation capacitor on a LM301 op amp; it was in a Schmitt trigger, with positive feedback.

It's also often widely mis-stated and mis-interpreted, in the same league as the Nyquist-Shannon samping theorem.